VType: Entering Text in a Virtual World

We have developed a software tool, VType, that enables a user wearing virtual reality gloves to enter text while in a virtual world. We present techniques to convert noisy data representing nger movements from the gloves into cleaner signals and determine the nger presses. Since each nger press corresponds to more than one symbol, we then use an algorithm for resolving ambiguity on such overloaded keyboard. We demonstrate that accurate text entry is possible using Vtype. INTRODUCTION Input in most present day virtual environments is characterized by the location and orientation information for users' head and possibly hands. Further, such information, generally acquired from trackers, is used only for coarse interactions , such as direction of viewing and location of hands relative to one another and with respect to the body. Even when gloves are used as a means for user input, they are typically used for very coarse gestures such as select, grasp, drop. In contrast, real life working environments are characterized by human adeptness with ne gestures. This work is an attempt to push the envelope on ne gesture recognition for meaningful tasks in virtual environments through relatively inexpensive hardware.